The Psychology of Hypothesis Testing:

-Traditional theory (now largely discredited): knowing a concept amounts to knowing how to classify some instance as an element of a certain category. It was argued that we classify instances on the basis of cues.

Ie: knowing the concept “spades” iis to know how to cards into spade and non-spade categories.

Concept formation:

-Conjunction of Cues: “Seven of Spades” is conjoined by cues “Seven” and “Spades”.

-Disjunction of Cues: “Earned Income” is disjoined by AND/OR the following: “wages, salaries, tips, commissions, etc.”

-Traditional view holds that we ordinarly learn cues by testing hypothesis. Ie: a child learning the word “dog” will get corrected (her evidence) when she points to a cow and say “dog”.

___________________________________

A STUDY:

Subjects were given a set of 81 cards, each with either:

- 1, 2, or 3 forms

- 1, 2, or 3 bordres

- Forms of all crosses, all squares or all circles.

- Color Green, Red, or Black.

The instructor had a category in mind. (ie: “black crosses”, or even just “black”.)

The subject was asked to determine the category by pointing to a card and asking if it fit in the category (YES) or if it did not fit in the category (NO).

This was called a selection experiment, because the experimenter selected the cards that he wanted. In reception experiments, the experimenter chooses cards randomly.

__________________________________________

One of the strategies that subjects used was simultaneous scanning. When the first YES answer was received, they noted everything about the card and used it as an initial hypothesis. (Ie: perhaps the card was “one red cross with two borders”). If the next YES is “two red crosses with three borders” he might conclude “red cross” as the category.

Successive Scanning is where the subject tests one hypothesis at a time. So, he would devise a hypothesis before testing (ie: Red) and then select red cards that differed on other attributes. (Ie: Red Cross and Red Circle). If Red Circle turns out to not be a part of the category, he might think the category is “Red Cross.”

If “Two Red Borders with Three Borders” gets a YES but “Two Red Borders with One Border” gets a NO, then the subject has good (certain?) reason to think borders are irrelevant (not part of the category title.) This is called conservative focusing.

-Another strategy is called focus gambling. If we take the hypothesis that gets a YES (two red crosses and three borders) and then make our next hypothesis differ in every regard except one (one red square with one border), then if we do get a YES, we have good reason to think the category is just “RED.” It was unlikely for the subject to get a YES, but if he does, then again, it is likely that his hypothesis is correct.

The view of concept learning that lay behind this work has since been largely discredited.

-Putnam points out that the concept of “water” has little to do with the features that we usually associate with water, such as being liquid, transparent, boil at 212 F etc. If a chemical had all of these features, but it had chemical formula XYZ and not water, then we would say that it was indeed XYZ because we take the chemical formula or water as the definitive word on what is and what is not water. Similarly, if a fruit looked and tasted like an orange, but grew on an apple tree, we wouldn’t call it an orange, we would call it a (very strange) apple.

-Wittgenstein pointed out also that the concept of games don’t have any definitive attributes, but rather each game shares some defining features with some other games. (Ie, for every fun game you can think of, there are also boring games, so not all games are fun). He said that games are united by a family resemblance.

Congruence Bias:

-Subjects tend to use direct hypothesis rather than indirect. Ie, suppose that there are two doors A and B with a reward behind it. They make a hypothesis, “Reward is behind door A” and they choose to look behind door A instead of door B.

-It appears that subjects ask questions designed to yeild yes answer if their favored hypothesis were true, even when these are not the best (most informative) questions

Experiment: Four cards are placed in an envelope, each with one letter (A B C) and one number (1 2 3). You are given a hypothesis, If there is a B, then there is a 2, and you are asked whether or not it is true for all four cards. Your task is to ask a question that could tell you whether the hypothesis is true or not.

A1 B1 C1

A2 B2 C2

A3 B3 C3

Note that the correct answer would include an inquiry about B1 and B3, as these are the only two cards that would falsify the hypothesis. However, many subjects only asked if there was a B2 in the envelope. Note that if they asked for any cards besides B1 and B3, the rule would still be true.

Two prescriptive heuristics

What is my dad’s favorite food?

Hypothesis: pizza

Possibilities: Corn, chili, pizza, sushi, avocado, rice.

A good prescriptive heuristic: How likely is a YES if I assume that my hypothesis is false?

In otherwords, how likely is my dads favorite food either corn, chili, sushi, salmon or salad? A good test is one that has a high probability of some answer if the hypothesis is true and a low probability if the hypothesis is false.

Another good prescriptive heuristic: Try to think of alternative hypothesis; then choose a test most likely to distinguish them - a test that will probably give different results depending on which is true.

A test could be to put pizza and a steak in the fridge and see what food my dad decides to take. (lol idunnolol…)

Congruence Bias: ...this whole damn section confused me. Research more about it later (pg 171-177) but here is a good tidbit of information:

-Reward vs no reward seems to cause different results in hypothesis testing.

-These experiments are probably analogous to real situations - for example, in classrooms - in which people are induced to work for extrinsic rewards such as grades. The reward may be effective in encouraging the work in question, but it may reduce the commitment to other valuable goals - such as satisfying one’s curiosity - that could otherwise motivate the same behavior. Once students find a way of getting the reward of grades, they may be less inclined to try other ways of thinking that might teach them something about the real world that would ultimately enable them to obtain rewards much more important to them than good grades.

Information Bias and the Value of Information

-Best test for a hypothesis is one that yields the most relevant information

-Patient has 80% chance of having globoma; if it isn’t glaboma it’s either popitis or flapemia. We can do one test and only one test: ET scan. ET scan will be positive if patient has popitis. Equally likely to be positive OR negative if patient has globoma. And will be negative if have flapemia. (Note that positive results will show for either popitis or globoma and negative will show for either flapemia or globoma. What should we do? - We should treat patient for Globoma and the test results has no influence on this decision. We treat the patient for Globoma because the likelihood (from symptoms and history) is so high. The test is irrelevant.

-Sometimes we want information for the sake of curiosity, but when we have other goals (ie, saving time and money), it would be irrational (and in fact useless) to go through with such a test.

-A test has some expected utility only if it can affect the action we would take. Subjects in experiments, however, show a bias in which they seek information even when it cannot affect action.

dafuq. more confusion pg 177-181

Thinking And Deciding